A novel method for landmark-based personal identification on panoramic dental radiographic and computed tomographic images. (December 2016)

• Record Type: Journal Article
• Authors: Fujimoto, Hideko; Hayashi, Takeshi; Iino, Morio
• Is Part Of: Journal of forensic radiology and imaging. Volume 7(2016)
• Page Start: 21
• Page End: 27
• Publication Date: 2016-12
• Subjects: Personal identification -- Panoramic CT -- Panoramic X-ray -- Landmark method -- Procrustes analysis -- Forensic odontology; Forensic radiography -- Periodicals; Magnetic resonance imaging -- Periodicals; Diagnostic imaging -- Periodicals; Diagnostic imaging; Forensic radiography; Magnetic resonance imaging; Periodicals; 614.1
• DOI: 10.1016/j.jofri.2016.10.002
• Languages: English
• ISSNs: 2212-4780

Abstract

Abstract: Personal identification based on dental work is a well-known and useful method for post-mortem identification. Recently, several odontologists have matched reconstructed post-mortem computed tomographic (CT) panoramic images of unidentified bodies with ante-mortem dental panoramic X-ray images for identification purposes. However, it is not always possible to obtain useful information from these panoramic X-ray and CT images. To obtain more accurate information for personal identification, we focused on the anatomical structures of the maxilla and mandible rather than dental work on the images. The aim of the present research is to develop a novel method for the personal identification of unidentified bodies, based on landmarks on the tooth sockets, which are not notably subject to post-mortem changes. We calculated the Procrustes distance and Pearson's correlation coefficients of the landmarks of two sample images, and used the results to correctly identify the subjects from among a pool of 100 candidates, without having to rely on information on dental work. We conclude that this method is a reliable tool and can be easily and rapidly applied for victim and missing person identification. Graphical abstract: Highlights: We performed personal identification using statistical shape analysis. We focused on the tooth sockets, because these were likely remained on any condition of the body. We divided each image of the jaws into 6 segments so that we could evaluate the positional relationships with high precision. We will be able to use the proposed method for DVI.